def test_error(self):
with paddle.static.program_guard(paddle.static.Program()):
weight_fp32 = paddle.data(
name='weight_fp32', shape=[1], dtype='float32')
# The input type must be Variable.
self.assertRaises(TypeError, F.prelu, x=1, weight=weight_fp32)
# The input dtype must be float16, float32, float64.
x_int32 = paddle.data(name='x_int32', shape=[2, 3], dtype='int32')
self.assertRaises(TypeError, F.prelu, x=x_int32, weight=weight_fp32)
# support the input dtype is float16
x_fp16 = paddle.data(name='x_fp16', shape=[2, 3], dtype='float16')
F.prelu(x=x_fp16, weight=weight_fp32)
def test_attr_tensor_API(self):
startup_program = Program()
train_program = Program()
with program_guard(train_program, startup_program):
fill_value = 2.0
input = paddle.data(name='input', dtype='float32', shape=[2, 3])
output = paddle.full_like(input, fill_value)
output_dtype = paddle.full_like(input, fill_value, dtype='float32')
place = paddle.CPUPlace()
if core.is_compiled_with_cuda():
place = paddle.CUDAPlace(0)
exe = paddle.static.Executor(place)
exe.run(startup_program)
img = np.array([[1, 2, 3], [4, 5, 6]]).astype(np.float32)
res = exe.run(train_program,
feed={'input': img},
fetch_list=[output])
out_np = np.array(res[0])
self.assertTrue(not (out_np - np.full_like(img, fill_value)).any(),
msg="full_like output is wrong, out = " +
str(out_np))
def test_api(self):
shape = [3, 4]
startup_program = Program()
train_program = Program()
with program_guard(train_program, startup_program):
x = paddle.data('X', shape)
# 'bool', 'float32', 'float64', 'int32', 'int64'
out1 = zeros_like(x)
out2 = zeros_like(x, np.bool)
out3 = zeros_like(x, 'float64')
out4 = zeros_like(x, 'int32')
out5 = zeros_like(x, 'int64')
place = fluid.CUDAPlace(
0) if core.is_compiled_with_cuda() else fluid.CPUPlace()
exe = fluid.Executor(place)
outs = exe.run(train_program,
feed={'X': np.ones(shape).astype('float32')},
fetch_list=[out1, out2, out3, out4, out5])
for i, dtype in enumerate(
[np.float32, np.bool, np.float64, np.int32, np.int64]):
self.assertEqual(outs[i].dtype, dtype)
self.assertEqual((outs[i] == np.zeros(shape, dtype)).all(), True)
def test_attr(self):
x = paddle.data(name='x', shape=[10, 10], dtype='float64')
def test_return_index():
result = paddle.unique(x, return_index=0)
self.assertRaises(TypeError, test_return_index)
def test_return_inverse():
result = paddle.unique(x, return_inverse='s')
self.assertRaises(TypeError, test_return_inverse)
def test_return_counts():
result = paddle.unique(x, return_counts=3)
self.assertRaises(TypeError, test_return_counts)
def test_axis():
result = paddle.unique(x, axis='12')
def test_dtype():
result = paddle.unique(x, dtype='float64')
self.assertRaises(TypeError, test_axis)
def test_with_zero_state(self):
mp = self.mp.clone()
sp = self.sp
rnn1 = self.rnn1
rnn2 = self.rnn2
exe = self.executor
scope = self.scope
x = np.random.randn(4, 16)
y1, (h1, c1) = rnn1(x)
with paddle.fluid.unique_name.guard():
with paddle.static.program_guard(mp, sp):
x_data = paddle.data(
"input", [-1, 16],
dtype=paddle.framework.get_default_dtype())
y, (h, c) = rnn2(x_data)
feed_dict = {x_data.name: x}
with paddle.static.scope_guard(scope):
y2, h2, c2 = exe.run(mp,
feed=feed_dict,
fetch_list=[y, h, c],
use_prune=True)
np.testing.assert_allclose(h1, h2, atol=1e-8, rtol=1e-5)
np.testing.assert_allclose(c1, c2, atol=1e-8, rtol=1e-5)
def static(self):
with paddle.static.program_guard(paddle.static.Program()):
x = paddle.data('X', self.shape, self.dtype)
out = paddle.var(x, self.axis, self.unbiased, self.keepdim)
exe = paddle.static.Executor(self.place)
res = exe.run(feed={'X': self.x}, fetch_list=[out])
return res[0]
def check_static_result_4(self, place):
paddle.enable_static()
with program_guard(Program(), Program()):
input_shape = (2, 3, 4, 5, 6)
pad = [1, 2, 1, 1, 3, 4]
mode = "circular"
input_data = np.random.rand(*input_shape).astype(np.float32)
x = paddle.data(name="x", shape=input_shape)
result1 = F.pad(x=x, pad=pad, mode=mode, data_format="NCDHW")
result2 = F.pad(x=x, pad=pad, mode=mode, data_format="NDHWC")
exe = Executor(place)
fetches = exe.run(default_main_program(),
feed={"x": input_data},
fetch_list=[result1, result2])
np_out1 = self._get_numpy_out(input_data,
pad,
mode,
data_format="NCDHW")
np_out2 = self._get_numpy_out(input_data,
pad,
mode,
data_format="NDHWC")
self.assertTrue(np.allclose(fetches[0], np_out1))
self.assertTrue(np.allclose(fetches[1], np_out2))
def test_api(self):
x_1 = paddle.data(shape=[None, 1, 4, 5], dtype='int32', name='x_1')
paddle.concat([x_1, x_1], 0)
input_2 = np.random.random([2, 1, 4, 5]).astype("int32")
input_3 = np.random.random([2, 2, 4, 5]).astype("int32")
x_2 = fluid.data(shape=[2, 1, 4, 5], dtype='int32', name='x_2')
x_3 = fluid.data(shape=[2, 2, 4, 5], dtype='int32', name='x_3')
positive_1_int32 = paddle.fill_constant([1], "int32", 1)
positive_1_int64 = paddle.fill_constant([1], "int64", 1)
negative_int64 = paddle.fill_constant([1], "int64", -3)
out_1 = paddle.concat(x=[x_2, x_3], axis=1)
out_2 = paddle.concat(x=[x_2, x_3], axis=positive_1_int32)
out_3 = paddle.concat(x=[x_2, x_3], axis=positive_1_int64)
out_4 = paddle.concat(x=[x_2, x_3], axis=negative_int64)
exe = paddle.static.Executor(place=paddle.CPUPlace())
[res_1, res_2, res_3,
res_4] = exe.run(paddle.static.default_main_program(),
feed={
"x_1": input_2,
"x_2": input_2,
"x_3": input_3
},
fetch_list=[out_1, out_2, out_3, out_4])
assert np.array_equal(res_1, np.concatenate((input_2, input_3),
axis=1))
assert np.array_equal(res_2, np.concatenate((input_2, input_3),
axis=1))
assert np.array_equal(res_3, np.concatenate((input_2, input_3),
axis=1))
assert np.array_equal(res_4, np.concatenate((input_2, input_3),
axis=1))
def run_static(self, use_gpu=False):
input = paddle.data(name='input', shape=[10, 10, 5], dtype='float32')
result0 = paddle.prod(input)
result1 = paddle.prod(input, axis=1)
result2 = paddle.prod(input, axis=-1)
result3 = paddle.prod(input, axis=[0, 1])
result4 = paddle.prod(input, axis=1, keepdim=True)
result5 = paddle.prod(input, axis=1, dtype='int64')
result6 = paddle.prod(input, axis=1, keepdim=True, dtype='int64')
place = paddle.CUDAPlace(0) if use_gpu else paddle.CPUPlace()
exe = paddle.static.Executor(place)
exe.run(paddle.static.default_startup_program())
static_result = exe.run(feed={"input": self.input},
fetch_list=[
result0, result1, result2, result3, result4,
result5, result6
])
expected_result = np.prod(self.input)
self.assertTrue(np.allclose(static_result[0], expected_result))
expected_result = np.prod(self.input, axis=1)
self.assertTrue(np.allclose(static_result[1], expected_result))
expected_result = np.prod(self.input, axis=-1)
self.assertTrue(np.allclose(static_result[2], expected_result))
expected_result = np.prod(self.input, axis=(0, 1))
self.assertTrue(np.allclose(static_result[3], expected_result))
expected_result = np.prod(self.input, axis=1, keepdims=True)
self.assertTrue(np.allclose(static_result[4], expected_result))
expected_result = np.prod(self.input, axis=1, dtype=np.int64)
self.assertTrue(np.allclose(static_result[5], expected_result))
expected_result = np.prod(
self.input, axis=1, keepdims=True, dtype=np.int64)
self.assertTrue(np.allclose(static_result[6], expected_result))
def test_error(self):
with paddle.static.program_guard(paddle.static.Program(),
paddle.static.Program()):
x = paddle.data(name='x', shape=[2, 2, 4], dtype='float32')
bool_x = paddle.data(name='bool_x', shape=[2, 2, 4], dtype='bool')
# The argument x shoule be a Tensor
self.assertRaises(TypeError, paddle.prod, [1])
# The data type of x should be float32, float64, int32, int64
self.assertRaises(TypeError, paddle.prod, bool_x)
# The argument axis's type shoule be int ,list or tuple
self.assertRaises(TypeError, paddle.prod, x, 1.5)
# The argument dtype of prod_op should be float32, float64, int32 or int64.
self.assertRaises(TypeError, paddle.prod, x, 'bool')
def test_x_dtype():
with paddle.static.program_guard(paddle.static.Program(),
paddle.static.Program()):
x = paddle.data(name='x', shape=[10, 10], dtype='float16')
result = paddle.unique(x)
self.assertRaises(TypeError, test_x_dtype)
def test_reflect_3():
input_shape = (1, 2, 3, 4, 5)
data = np.random.rand(*input_shape).astype(np.float32)
x = paddle.data(name="x", shape=input_shape)
y = F.pad(x, pad=[1, 1, 1, 1, 2, 3], value=1, mode='reflect')
place = paddle.CPUPlace()
exe = Executor(place)
outputs = exe.run(feed={'x': data}, fetch_list=[y.name])
def test_type():
# dtype must be float32, float64, int8, int32, int64.
x2 = np.arange(image_shape[0] * image_shape[1] * image_shape[2] *
image_shape[3]).reshape(image_shape) / 100.
x2 = x2.astype('float16')
x2_var = paddle.data(name='x2',
shape=[3, 2, 4, 5],
dtype='float16')
paddle.flatten(x2_var)
def test_attr(self):
x = paddle.data(name='x', shape=[10, 10], dtype='float64')
y = paddle.data(name='y', shape=[10, 10], dtype='float64')
def test_rtol():
result = paddle.allclose(x, y, rtol=True)
self.assertRaises(TypeError, test_rtol)
def test_atol():
result = paddle.allclose(x, y, rtol=True)
self.assertRaises(TypeError, test_atol)
def test_equal_nan():
result = paddle.allclose(x, y, equal_nan=1)
self.assertRaises(TypeError, test_equal_nan)
def test_out2(self):
with paddle.static.program_guard(paddle.static.Program(),
paddle.static.Program()):
x = paddle.data('x', shape=[-1, 2], dtype='float64')
index = paddle.data('index', shape=[-1, 1], dtype='int32')
axis = paddle.data('axis', shape=[1], dtype='int32')
out = paddle.gather(x, index, axis)
place = paddle.CPUPlace()
exe = paddle.static.Executor(place)
x_np = np.array([[1, 2], [3, 4], [5, 6]]).astype('float64')
index_np = np.array([1, 1]).astype('int32')
axis_np = np.array([1]).astype('int32')
result, = exe.run(
feed={"x": x_np,
"index": index_np,
'axis': axis_np},
fetch_list=[out])
expected_output = gather_numpy(x_np, index_np, axis_np)
self.assertTrue(np.allclose(result, expected_output))
def test_static(x_np, y_np, p=2.0, epsilon=1e-6, keepdim=False):
prog = paddle.static.Program()
startup_prog = paddle.static.Program()
place = fluid.CUDAPlace(0) if paddle.fluid.core.is_compiled_with_cuda(
) else fluid.CPUPlace()
with paddle.static.program_guard(prog, startup_prog):
x = paddle.data(name='x', shape=x_np.shape, dtype=x_np.dtype)
y = paddle.data(name='y', shape=y_np.shape, dtype=x_np.dtype)
dist = paddle.nn.layer.distance.PairwiseDistance(
p=p, epsilon=epsilon, keepdim=keepdim)
distance = dist(x, y)
exe = paddle.static.Executor(place)
static_ret = exe.run(prog,
feed={'x': x_np,
'y': y_np},
fetch_list=[distance])
static_ret = static_ret[0]
return static_ret
def test_l2_loss(self):
student_main = fluid.Program()
student_startup = fluid.Program()
with fluid.program_guard(student_main, student_startup):
input = paddle.data(name="image", shape=[None, 3, 224, 224])
conv1 = conv_bn_layer(input, 8, 3, "conv1")
conv2 = conv_bn_layer(conv1, 8, 3, "conv2")
student_predict = conv1 + conv2
teacher_main = fluid.Program()
teacher_startup = fluid.Program()
with fluid.program_guard(teacher_main, teacher_startup):
input = paddle.data(name="image", shape=[None, 3, 224, 224])
conv1 = conv_bn_layer(input, 8, 3, "conv1")
conv2 = conv_bn_layer(conv1, 8, 3, "conv2")
sum1 = conv1 + conv2
conv3 = conv_bn_layer(sum1, 8, 3, "conv3")
conv4 = conv_bn_layer(conv3, 8, 3, "conv4")
sum2 = conv4 + sum1
conv5 = conv_bn_layer(sum2, 8, 3, "conv5")
teacher_predict = conv_bn_layer(conv5, 8, 3, "conv6")
place = fluid.CPUPlace()
data_name_map = {'image': 'image'}
merge(teacher_main, student_main, data_name_map, place)
merged_ops = []
for block in student_main.blocks:
for op in block.ops:
merged_ops.append(op.type)
with fluid.program_guard(student_main):
distill_loss = l2_loss('teacher_conv6_bn_output.tmp_2',
'conv2_bn_output.tmp_2', student_main)
loss_ops = []
for block in student_main.blocks:
for op in block.ops:
loss_ops.append(op.type)
self.assertTrue(set(merged_ops).difference(set(loss_ops)) == set())
self.assertTrue(
set(loss_ops).difference(set(merged_ops)) ==
{'reduce_mean', 'square', 'elementwise_sub'})
def api_case(self, axis=None, keepdim=False):
out_ref = ref_logsumexp(self.x, axis, keepdim)
with paddle.static.program_guard(paddle.static.Program()):
x = paddle.data('X', self.shape)
out = paddle.logsumexp(x, axis, keepdim)
exe = paddle.static.Executor(self.place)
res = exe.run(feed={'X': self.x}, fetch_list=[out])
self.assertTrue(np.allclose(res[0], out_ref))
paddle.disable_static(self.place)
x = paddle.to_variable(self.x)
out = paddle.logsumexp(x, axis, keepdim)
self.assertTrue(np.allclose(out.numpy(), out_ref))
paddle.enable_static()
